Synthetic Data Vault (SDV)

The Synthetic Data Vault (SDV) is a Python library for creating tabular synthetic data. Users train their own generative model on real data, choosing among algorithms designed for single-table, sequential or multi-table (relational) data, and then create any amount of synthetic data on demand.

SDV evaluates the statistical quality of synthetic data, diagnoses problems and produces visualisations comparing synthetic and real data. Synthesizers can be customised with constraints, preprocessing settings and anonymisation options, and the library runs on-premises on standard CPUs.

SDV Community is published on GitHub by DataCebo under the Business Source License 1.1 (version 2.0.0 in October 2026); SDV Enterprise, available to licensed users, adds features for large, complex sets of interconnected tables.